elfman
Mechanical
- Oct 21, 2003
- 79
I will try to elaborate on the specifics of what I am trying to do, as hopefully someone will have some keen insight that will help. I actually had a smaller version of this posted to the Vibs and Acoustics forum, but it was deleted.
We have a bearing structure that is a single shaft with races ground into each end producing a straddle bearing. One end of the shaft has a flange on which we mount a rotor and a relatively large mass. The CoM of the large mass is close to centered between the races. Rather than building a complex, long running model, I proceeded with the knowledge that resonance, or natural frequency(fn) is simply sqrt(k/m). So, I removed all the components of the assembly that I don't intend to change and said, yes they will affect the fn, but they aren't changing from design to design. I cannot change the mass of the system, as it is fixed. I want to lower the fn to below - preferably well below the running frequency ~ 150-180Hz so I need to change the stiffness. Stiffness is simply a force-deflection thing. My model consists of two parts, the bearing shaft with simple, cylindrical supports centered in each of the bearing races, and the part that interfaces the remaining masses to the flange on the shaft as well as the masses (modeled as point masses) of the other two components located at their respective CoM and attached to their respective interfaces. I have run numerous simulations trying to correlate the deflection of the CoM to the resonance data I have from actual working assemblies. When I revised the design and built & tested some of my predicted 'lower resonance' assemblies (In Workbench, the predicted deflections matched the predicted deflections of the other low resonance assemblies) I found that if I changed the resonance it wasn't enough to make a measurable difference. I am not doing any rotational analysis in ANSYS, just stiffness and all this was done in ANSYS Workbench v10sp1.
I have since tried the Modal analysis in WB v11.0 and haven't seen much correlation there either. I am not trying to predict the exact resonance, just find something that correlates so that I can quickly iterate through some designs. The goal here is to quickly identify some other lower resonance designs without building a bunch of hardware. Any ideas, suggestions, and or Louisville Sluggers upside the back of my head would be appreciated.
Thank you.
We have a bearing structure that is a single shaft with races ground into each end producing a straddle bearing. One end of the shaft has a flange on which we mount a rotor and a relatively large mass. The CoM of the large mass is close to centered between the races. Rather than building a complex, long running model, I proceeded with the knowledge that resonance, or natural frequency(fn) is simply sqrt(k/m). So, I removed all the components of the assembly that I don't intend to change and said, yes they will affect the fn, but they aren't changing from design to design. I cannot change the mass of the system, as it is fixed. I want to lower the fn to below - preferably well below the running frequency ~ 150-180Hz so I need to change the stiffness. Stiffness is simply a force-deflection thing. My model consists of two parts, the bearing shaft with simple, cylindrical supports centered in each of the bearing races, and the part that interfaces the remaining masses to the flange on the shaft as well as the masses (modeled as point masses) of the other two components located at their respective CoM and attached to their respective interfaces. I have run numerous simulations trying to correlate the deflection of the CoM to the resonance data I have from actual working assemblies. When I revised the design and built & tested some of my predicted 'lower resonance' assemblies (In Workbench, the predicted deflections matched the predicted deflections of the other low resonance assemblies) I found that if I changed the resonance it wasn't enough to make a measurable difference. I am not doing any rotational analysis in ANSYS, just stiffness and all this was done in ANSYS Workbench v10sp1.
I have since tried the Modal analysis in WB v11.0 and haven't seen much correlation there either. I am not trying to predict the exact resonance, just find something that correlates so that I can quickly iterate through some designs. The goal here is to quickly identify some other lower resonance designs without building a bunch of hardware. Any ideas, suggestions, and or Louisville Sluggers upside the back of my head would be appreciated.
Thank you.